OMCT Simulation for AOD1B RL04

Short-term ocean mass re-distributions as simulated with the global Ocean Model for Circulation and Tides (OMCT; Thomas 2002) are applied in standard GRACE gravity field processing in order to reduce short-term ocean mass anomalies. The simulation described here is applied in release 04 of the Atmosphere Ocean De-aliasing Product (AOD1B RL04; Flechtner, 2005).

The model run is performed on a horizontal 1.875° x 1.875° grid with 13 vertical layers and a time-step of 30 minutes. Starting from an initial model state based on climatological simulations using wind stresses (Hellerman and Rosenstein, 1983) and mean sea surface temperatures and salinities (Levitus, 1982), the model has been forced by 6-hourly resolved ERA-40 reanalysis data (Uppala et al., 2006) covering the period 1958 to 2000. Subsequently, operational ECMWF analysis data are used to force the model. Momentum transfers due to winds and atmospheric pressure, heat fluxes indicated by 2m-temperatures and atmospheric freshwater fluxes due to precipitation and evaporation have been taken into account. Due to the omission of continental freshwater fluxes and the existence of large-scale trends in the precipitation and evaporation data, the total ocean mass is artificially held constant throughout the time by means of adding a spatially homogeneous layer of mass at each time-step (Dobslaw and Thomas, 2007b). Lunisolar tidal dynamics are not included within this particular model run.

Simulations are performed daily on a routine basis with about 4 days latency, mainly caused by availability restrictions of atmospheric forcing data. Relevant ocean state parameters and diagnosed fields are stored every 6 hours, on example of ocean bottom pressure anomalies is displayed in Figure 1.

Simulated mean sea surface heights (Figure 2 ) are closely related to mean surface currents via the geostrophic relation.

Topography gradients indicating high current velocities are apparent in the Southern Oceans connected to the Antarctic Circumpolar Current (ACC, Figure 3) or along the western boundary currents Kuroshio (North Pacific) and Gulf Stream (North Atlantic), whose transport is in particular slightly underestimated within OMCT.

The total variability of sea surface heights generally reflects the ocean's response to atmospheric pressure loading, to changes in the density distribution of the water column affecting the steric component of the sea surface height as well as to mass changes of the local water column represented by ocean bottom pressure (Figure 4). While atmospheric loading is most pronounced in middle latitudes, sterically induced sea surface height variability dominates in lower latitudes and in the vicinity of the western boundary currents, while changes in ocean mass are primarily concentrated in the North Pacific as well as in various regions of the Southern Oceans.

Sea surface temperatures (Figure 5) reveal typical zonally oriented patterns with highest temperatures in the tropics. Differences with respect to the WOA2001 climatology (Conkright et al., 2002) are generally below 2°C, highest deviations occur within the western boundary currents, where the reproduction of the exact location of the strong lateral temperature gradients is prohibited by the spatial resolution of the model. Mean sea surface salinites are coupled to the WOA2001 climatology in OMCT, and the relaxation time-scale of 38 days damps down seasonal variations. Strongest variability is identified in partially ice-covered regions and in the tropics, where strong precipitation events occur frequently.

In its present configuration, OMCT is forced by atmospheric pressure obtained from ECMWF. Periodic fluctuations in atmospheric temperature and, consequently, pressure called atmospheric tides (cf. Chapman and Lindzen, 1970), induce a corresponding response is excited in the ocean, leading to an additional ocean tide. While the diurnal atmospheric tide is fully retained in the model output (Figure 6), semidiurnal variability has been largely removed from the forcing data by means of a correction model (Ray and Ponte, 2003). Remaining semidiurnal variability, which aliases into a standing wave pattern due to the 6-hourly temporal resolution of the forcing and output fields, is significantly lower than the original S2(p) variability (see Dobslaw and Thomas, 2005).